Every Post Scheduled for April 18th Was Published Simultaneously at 12:00:00 AM — 2.3 Million Posts Hit the Platform in Under One Second, Collapsing Feed Infrastructure for 94 Minutes

At exactly midnight, MetaCity's scheduling queue — which holds posts queued by creators for staggered publication throughout the day — executed all pending jobs simultaneously. In under one second, 2.3 million posts were pushed to the platform. The feed infrastructure, which is designed for a maximum burst rate of approximately 40,000 posts per second, collapsed within 3 seconds. For 94 minutes, no feed on MetaCity loaded correctly. Some users saw a single post repeated infinitely. Some users saw a completely empty feed. Some users saw posts from accounts they had blocked years ago. The scheduling queue has been suspended pending investigation. Creators whose scheduled posts published at midnight have been told the posts 'cannot be unscheduled retroactively.'

MetaCity's post scheduling system processes approximately 2.3 million scheduled jobs per day. Creators use it to queue posts for optimal publication times — early morning slots for maximum morning traffic, specific hours for audience-targeted content, coordinated release windows for multi-creator campaigns. The scheduling system is supposed to release these posts in a continuous staggered flow throughout the day, smoothing the publication load across 24 hours and ensuring that each post has a clean, uncrowded moment in the feed when it publishes. At 12:00:00 AM on April 18th, the scheduling system did not stagger anything. A configuration error in the queue processing daemon — introduced during a routine maintenance window earlier in the evening — caused the job dispatcher to interpret all pending jobs as having a scheduled time of midnight and execute them simultaneously. In under one second, 2.3 million posts were sent to the platform's write pipeline.

The write pipeline is sized for a maximum burst capacity of approximately 40,000 posts per second. It received 2.3 million in under one second — a load approximately 57 times its designed maximum. The pipeline collapsed within 3 seconds. The failure propagated outward: the feed assembly service, which reads from the write pipeline's output, received malformed data and began serving corrupted feed states to users. The failure modes varied by user segment depending on which feed server each user was assigned to. Users on servers that received an overflow burst saw a single post from the 2.3 million repeated infinitely in their feed — the last successfully written entry before the pipeline entered its error state. Users on servers that lost connection to the write pipeline entirely saw an empty feed. Users on servers that entered a cache recovery state saw cached feed data from up to 72 hours prior — posts from accounts they had blocked, content from categories they had muted, and in several cases, content from accounts that had since been deactivated.

2.3 Million Posts, Zero Seconds Apart, One Very Bad Morning

The 94-minute outage affected all feed types: main feed, explore feed, district public feeds, creator profile feeds, and live stream discovery. Live streams themselves remained accessible — the streaming infrastructure runs on a separate pipeline — but the discovery surface through which most users find new streams was down for the full duration. Three creators who had live shows scheduled for the early morning hours — an overnight music set, a cooking stream, a news roundup — launched into feeds that no one could see. All three continued their shows. MetaCity's infrastructure team restored basic feed functionality at 1:34 AM. Full feed normalization took until approximately 3:00 AM. The 2.3 million posts that published at midnight remained published — their timestamps read 12:00:00 AM, all 2.3 million of them, in a feed that was broken for the first 94 minutes after they went live.

The creator impact is substantial and has no immediate remedy. MetaCity's post scheduling system was the primary tool used by professional creators to manage their publication cadence — the carefully considered timing that determines whether a post reaches its intended audience at a moment of peak engagement or gets buried. All 2.3 million posts that were scheduled for April 18th published simultaneously at midnight instead of at their intended times. MetaCity's response to creator support requests asking about rescheduling or deletion has been consistent: 'Posts that have been published cannot be unscheduled retroactively.' The scheduling queue remains suspended. Creators cannot queue new posts. The posts that published at midnight are in the feed, timestamped incorrectly, timed incorrectly, performing incorrectly, and permanent. MetaCity has not announced a compensation mechanism. It has not announced when the scheduling queue will be restored. It has not explained what will happen to posts that were scheduled for future dates beyond April 18th that are now stranded in the suspended queue.